Experts talk about AI intellectual property: whether AI creations are protected by copyright and how they are regulated

At present, the cross-border application of artificial intelligence (AI) large models is showing a blowout trend, and generative artificial intelligence (AIGC) is emerging in the field of content creation. As robots can pass the Turing test, are granted legal status, and ChatGPT is popular out of the circle, these all mark the arrival of the AI era.

While the progress of AI technology has promoted profound changes in the production and life of the whole society, it has also brought new intellectual property challenges: What changes will AI bring to the intellectual property industry? Will AI make it easier to infringe intellectual property rights? How is the intellectual property rights of AIGC products defined?

On September 19-20, the 12th China Intellectual Property Annual Conference with the theme of "Intellectual Property Support for Comprehensive Innovation" was held in Jinan, Shandong Province, and a sub-forum on "Intellectual Property Escort, Leading the New Track of Artificial Intelligence" was set up at the annual meeting, where experts and scholars from academia and industry conducted in-depth exchanges and discussions on AI intellectual property topics. After the meeting, a number of experts were interviewed by The Paper (www.thepaper.cn) reporters on the above issues and shared their views on the above issues.

AI can help identify and prevent IP infringement

Cui Guobin, a professor at Tsinghua University Law School and director of the Intellectual Property Law Research Center, said in an interview with The Paper that the development of new technologies will always bring challenges to the intellectual property protection system from all aspects, and this process is endless.

First of all, new technologies have expanded the scope of potential intellectual property protection objects, such as whether and how new objects such as virtual character images, artificial intelligence creations, and big data collections should obtain intellectual property protection, which is a hot issue at present.

Second, new technologies lead to the emergence of new business models, and new exploits of works will emerge, and IP law needs to be adjusted accordingly. For example, the legality of copyright in NFT trading behaviors involving works and data use behaviors in the process of artificial intelligence training requires timely responses from decision makers.

Third, new technologies may also change the dynamic balance between IP infringement and prevention, and IP law needs to constantly adjust the rights and obligations established by right holders, intermediaries and the public, redistribute the cost of infringement prevention, and achieve a new dynamic balance. For example, blockchain, content filtering, artificial intelligence and other technologies can help identify and prevent intellectual property infringement in many fields, which may lead legislators to allow service providers in specific fields to adopt relevant technologies to more effectively prevent infringement.

Finally, new technologies also enable more open innovation models that do not rely on strict intellectual property protection. For example, the open source model in the software field has attracted more and more attention from the whole society. For policymakers, how to promote the development of the open innovation model and coordinate its relationship with the IP protection mechanism has become a new issue.

The copyright regulation of AI creations is facing challenges

Speaking of the biggest challenge brought by new technologies to intellectual property rights, Cui Guobin believes that it is the copyright regulation of artificial intelligence: "The copyright law needs to immediately answer the extent to which AI creations should be protected by copyright, whether the use of data in works by AI developers constitutes fair use, and the duty of care of AI developers to prevent infringement of outputs." ”

The development of new technologies will inevitably make the public's communication capacity more and more powerful, which means that infringement activities are more decentralized and more difficult to prevent. As a result, it is more difficult to enforce intellectual property rights and enforce them. Cui Guobin said that the society's response to this problem is also multifaceted, so he put forward four suggestions:

First, legislators are likely to strengthen the duty of care of some middlemen, requiring them to more actively prevent or stop infringing activities by public users, or to develop appropriate business models to more actively seek authorization from IP owners on behalf of public users.

Second, society will also develop more effective technical measures, such as watermarking, encryption, filtering and other technologies, to prevent public infringement.

Third, rights holders will also strive to adapt to the new environment and strive to reduce the transaction costs of public authorization, thereby squeezing the profit margins of infringing activities. Even right holders will develop more business models that do not rely on intellectual property protection, such as free works but charging for services, works free for the public but charging for advertising, etc.

Finally, for some core areas where IP protection must be strengthened, legislators may strengthen legal liability for infringement and enhance the deterrent effect of the law, such as strengthening the infringement damages system, pursuing punitive damages, and pursuing more administrative or criminal liability.

Inclusive and prudent supervision of intellectual property rights in artificial intelligence should be implemented

Are AI generators works? Some people have suggested that since artificial intelligence is not a natural person and is not a qualified subject of copyright, AIGC products are not works; There is also a view that the copyright of AIGC products should belong to the programmer.

In this regard, Yi Jiming, director of the International Intellectual Property Research Center of Peking University, told The Paper that in his opinion, artificial intelligence products should be recognized as works. The creation should not be limited to the subject, but should be given the corresponding rights from the objective creation of the work itself, and AI should be given certain copyright. In Yi Jiming's view, the company, natural persons and AI constitute the main body of AIGC products. Although AI is ancillary therein, it does not affect whether it can constitute a work within the meaning of copyright law.

Yi Jiming pointed out that the current legal system for AI intellectual property protection in mainland China is not perfect enough, and suggested that the intellectual property protection principle for intelligent creation achievements be established, and from this principle, specific laws and regulations should be extended to form general provisions of intellectual property innovation principles.

"In addition, in the supervision of science and technology, inclusive and prudent supervision should be adopted, because science and technology is constantly innovating, innovation will always involve areas unknown to human beings, and when we cannot fully grasp it, there should be a certain degree of inclusiveness." Yi Jiming said that the supervision of intellectual property rights in the field of AI should not be too strict, and it will truly play a stimulating role in innovation.

Speaking of the regulatory advice on intellectual property rights in the field of AI, Cui Guobin said that in the current international competition for AI technology research and development, the intellectual property protection system plays an important role. On the whole, China already has a relatively complete intellectual property protection framework for technological innovation achievements. The protection of trade secrets, patents and software copyrights should be the most important part of this, and the mainland has also established corresponding systems. Next, relevant law enforcement and judicial work need to be strengthened so that the property rights protection provided by the law can be implemented, so that investors can have sufficient confidence in recovering their return on investment.

Cui Guobin said that in the field of patents, we should focus on improving the quality of patent authorization and avoiding the granting of low-quality patents. Overall, AI is a patent-intensive technology field, and companies are vulnerable to patent jungles. In this area, the Patent Law does not require additional preferential treatment. Courts or enforcers should avoid lowering patent licensing standards, and avoid combining too many traditional technologies with AI scenarios and patenting, hindering normal technology competition.

In addition, Cui Guobin also said that there is still much room for improvement in the mainland's trade secret protection system. For AI companies, technical secrets and secret data rely on this protection, and it is the most important property rights protection system. The limited provisions of the existing Anti-Unfair Competition Law on trade secret protection cannot meet the actual needs.